Shared genetic architecture between Parkinson’s disease and self-reported sleep-related traits implicates the MAPT locus on chromosome 17

Abstract

Study Objectives

Parkinson’s disease (PD) is a neurodegenerative disorder characterized by motor and non-motor symptoms. Among the latter, sleep disturbances are particularly common and include insomnia, obstructive sleep apnea (OSA), and excessive daytime sleepiness. Here, we investigated the shared genetic architecture between PD and sleep-related traits to uncover biological pathways that underpin this relationship.

Methods

We analyzed genome-wide association study (GWAS) summary statistics for PD (~37.7 K cases, ~18.6 K proxy cases, ~1.4 M controls) and eight self-reported sleep-related traits (each with n > 300 000): ease of getting up, chronotype (morningness), napping, insomnia, OSA, snoring, daytime dozing, and sleep duration. Genetic correlations were estimated using Linkage Disequilibrium (LD) score regression, and GWAS-pairwise analysis was used to identify genomic segments harboring shared causal variants. We then mapped these variants to protein-coding genes.

Results

We observed a genome-wide genetic correlation between PD and daytime dozing (p < .05). A separate, local-level analysis identified six genomic regions harboring shared causal variants between PD and other sleep-related traits (primarily ease of getting up and napping). The most statistically significant of these local associations was observed at a single locus on chromosome 17, which contains the majority of mapped protein-coding genes, including ARHGAP27, PLEKHM1, CRHR1, and MAPT. These genes are implicated in neurodegeneration and circadian rhythm regulation.

Conclusions

These findings suggest that the MAPT locus, beyond its established role in PD, may also contribute to sleep–wake regulation via shared biological pathways, including tau pathology, stress response, and chromatin remodeling. Our results highlight sleep disturbances as a potential early marker or risk factor of PD.

Statement of Significance

Sleep disturbances are one of the most common non-motor symptoms of Parkinson’s disease (PD), sometimes manifesting years before motor symptoms emerge, yet the genetic factors underlying this relationship remain poorly understood. Our study identifies shared risk loci in six genomic segments, highlighting a key region on chromosome 17 linked to both neurodegeneration and sleep–wake regulation. This work fills a critical gap by revealing specific genes and biological pathways that may drive both disease onset and sleep disruption. Future research should explore these pathways in diverse populations and investigate their potential as early biomarkers for interventions to screen or prevent PD.

Introduction

Parkinson’s disease (PD) is a progressive neurodegenerative disorder that affects ~2% of individuals over 60 years old [1]. While PD is primarily characterized by motor symptoms such as bradykinesia, tremor, and rigidity, it also manifests through a broad range of non-motor symptoms, including cognitive impairment, mood disorders, and sleep disturbances. The disease is strongly associated with the progressive degeneration of dopaminergic neurons in the substantia nigra, driven by the pathological accumulation of alpha-synuclein aggregates [2]. Notably, these pathological processes have also been implicated in regulating sleep–wake cycles [3].

Sleep disturbances are among the most prevalent non-motor symptoms in PD, affecting between 60% and 98% of individuals with the disorder [4, 5]. Common sleep-related symptoms in PD include insomnia, excessive daytime sleepiness, restless legs syndrome (RLS), nocturnal akinesia, rapid eye movement sleep behavior disorder (RBD), circadian rhythm disruptions, and obstructive sleep apnea (OSA) [6, 7]. The prodromal presence of RBD has been associated with a more severe motor and non-motor PD subtype, suggesting a potential disease-modifying role of this parasomnia [8]. A large prospective, population-based study demonstrated that subjectively poor sleep quality and shorter sleep duration were associated with an increased risk of incident Parkinsonism within the first 2 years of follow-up [9]. These findings underscore the importance of investigating sleep disturbances and their potential shared genetic basis with PD, as they may serve as early biomarkers of disease onset, particularly in individuals who are genetically susceptible.

Genetic factors are known to contribute to individual differences in the etiology of both PD and sleep-related traits. Twin and family studies have reported moderate to high heritability estimates for sleep phenotypes, including sleep duration (0.46 [10], insomnia (0.38–0.59 [11]), and napping 0.65 [12]). Likewise, in individuals of European ancestry, the heritability for PD has been estimated at 0.34 [13]. Genome-wide association studies (GWAS) have identified more than 134 genetic risk loci associated with PD [14], and similarly large numbers of loci have been linked to sleep-related traits [15], with one GWAS for insomnia reporting up to 554 risk loci [16].

Although emerging evidence suggests that the interplay between neurodegeneration, neurotransmitter imbalances, and genetic factors may underlie the connection between PD and sleep, the genetic architecture of this relationship remains poorly understood. In this study, we leveraged GWAS summary statistics for PD and eight sleep-related traits (OSA, daytime dozing, ease of getting up, sleep duration, napping, insomnia, morningness, and snoring) to identify shared genetic loci and map associated variants to protein-coding genes. Through this exploratory approach, we aimed to elucidate the genetic basis of the relationship between sleep and PD, providing new insights into the underlying biological pathways.

Methods

GWAS summary statistics for Parkinson’s disease

We leveraged GWAS summary statistics for PD representing individuals of European ancestry obtained from a previously published large-scale GWAS meta-analysis [17]. GWAS meta-analyses utilize data from multiple independent cohorts to substantially increase sample size and, therefore, statistical power to detect genetic variants associated with a specific phenotype. Accordingly, the resulting PD GWAS meta-analysis dataset comprised 37 688 clinically diagnosed PD cases sourced from independent cohorts, including Nalls et al. 2014 (n = 13 708), the International Parkinson’s Disease Genomic Consortium (IPDGC; n = 8036), the Parkinson’s Disease Web-Based Study (n = 6476), IPDGC-NeuroX (n = 5851), 23andMe (n = 2448), and the Systems Genomics of Parkinson’s Disease consortium (n = 1169). The UK Biobank contributed 18 618 proxy cases (i.e. individuals with a first-degree relative diagnosed with PD). The control group consisted of 1 417 791 individuals, totaling 1 474 097 participants. The meta-analysis was performed using a fixed-effects model implemented in METAL [18]. Access to the 23andMe cohort was obtained through the appropriate application process (https://research.23andme.com/dataset-access/) and governed by an institutional data transfer agreement. Specific sample sizes per cohort, as well as further details regarding GWAS design and quality control procedures, can be found in the original publication [17].

GWAS summary statistics for self-reported sleep-related traits

We used GWAS summary statistics from individuals of European ancestry for eight self-reported sleep-related traits: OSA (UK Biobank, n = 523 366), snoring (UK Biobank, n = 408 317), insomnia (UK Biobank and 23andMe, n = 2 365 010), daytime dozing (UK Biobank, n = 386 548), ease of getting up (UK Biobank, n = 385 949), sleep duration (UK Biobank, n = 384 317), napping (UK Biobank, n = 386 577), and morningness (UK Biobank, n = 345 552). Participants with missing or ambiguous responses (e.g. “prefer not to answer” or “I don’t know”) were excluded. Summary statistics were processed using strict quality control criteria for low imputation quality score (INFO < 0.6) and for minor allele frequency (MAF < 0.01). Full details of the cohorts and quality control procedures are available in the original publications [15, 19, 20].

Genetic correlations

We estimated genome-wide genetic correlations between PD and each of the eight sleep-related traits using linkage disequilibrium score regression (LDSC) [21]. This approach quantifies the extent to which genetic effects are shared across traits by correlating GWAS effect sizes across the genome. To account for multiple comparisons, we applied a Bonferroni correction, setting the significance threshold at p < .05/8 (number of sleep-related traits tested).

Pairwise analysis of GWAS

To identify genomic regions harboring variants that influence both PD and at least one sleep-related trait, we conducted an exploratory GWAS-pairwise (GWAS-PW) analysis with no loci of interest defined a priori. This method divides the genome into ~1700 independent regions based on linkage disequilibrium patterns and calculates the posterior probabilities of four models: (i) association with PD only; (ii) association with a sleep-related trait only; (iii) shared association with both PD and a sleep-related trait via common causal variants; and (iv) distinct causal variants for each trait [22]. Genomic segments with a posterior probability of association (PPA) > 0.5 [23–26] under model (iii) were prioritized for downstream analyses, in line with prior work.

Multivariate analysis of genomic annotation

We used MAGMA v1.08 [27] within the Functional Mapping and Annotation platform v1.6.0 [28] to further investigate the genomic segments identified through GWAS-PW. Specifically, we conducted gene-based and gene-set analyses to map variants in these regions to protein-coding genes jointly associated with PD and sleep-related traits. For the gene-based association tests, statistical significance was defined using Bonferroni correction for the total number of protein-coding genes tested per trait (p < .05/19 427) [29].

Regarding gene-set analyses, for the PD, we applied a Bonferroni correction based on the total number of gene-sets tested (n = 17 008). While for ease of getting up, napping, and sleep duration, which were the prioritized sleep-related traits based on the highest statistical significance of shared genomic segments with PD (GWAS-PW PPA_Model3 > 0.8) and the highest gene density within shared segments, we performed a targeted analysis restricted to the shared genomic segments with PD identified via GWAS-PW (PPA_Model3 > 0.5). Accordingly, gene-set significance for ease of getting up, napping, and sleep duration traits was corrected for the number of genes in the shared genomic segments (p < .05/n) [29].

Protein–protein interaction network analysis

We performed protein–protein interaction (PPI) network analysis using the STRING database v12.0 [30] to examine biological interactions in the most relevant shared loci. We included significant protein-coding genes mapped by MAGMA. Interaction scores were computed based on multiple sources of evidence, including experimental data, database annotations, text mining, and co-expression. We retained medium and high-confidence interactions with a score ≥ 0.4. We also applied the Markov Cluster Algorithm (MCL) with an inflation parameter of three to identify potential functional modules. Network enrichment was assessed by comparing the observed number of edges with the expected number given the size of the input set, using STRING’s built-in enrichment test.

Results

We estimated genome-wide genetic correlations between self-reported sleep-related traits and PD using linkage disequilibrium score regression. Daytime dozing showed a nominally significant negative genetic correlation with PD (, p-value = .0485). However, after correcting for multiple testing, no genetic correlations between PD and the sleep-related traits reached statistical significance (Figure 1; Supplementary Table S1).

Figure 1.

Genetic correlations between PD and sleep-related traits. Forest plot displaying the genome-wide genetic correlations () estimated using LDSC. The central markers represent the point estimates for the genetic correlation between PD and each sleep-related trait. Error bars indicate the 95% confidence intervals. The dashed vertical line at 0.0 represents no genetic correlation. Traits are ordered by the magnitude of the correlation.

To further investigate the shared genetic architecture at a local level, we performed GWAS-PW analyses (Supplementary Table S2). We identified six genomic segments that harbored shared causal variants between PD and at least one self-reported sleep-related trait (Table 1, Figure 2). Using MAGMA, we mapped genetic variants within these segments to 25 protein-coding genes (Supplementary Tables S3 and S4). Among the self-reported sleep-related traits, ease of getting up had the highest number of shared genomic segments with PD (four segments), followed by morningness and sleep duration (two segments each) (Supplementary Table S5).

Table 1.

Summary of significant genomic segments with shared causal variants between PD and sleep-related traits

Chromosome	Start of segment (bp)	End of segment (bp)	Sleep-related trait	PPA Model 1 (PD only)	PPA Model 2 (sleep only)	PPA Model 3(shared)	PPA Model 4(distinct)
5	87 390 784	88 890 359	Ease of getting up	0.036	0.000	0.502	0.461
5	87 390 784	88 90 359	Sleep duration	0.008	0.045	0.811	0.127
5	136 376 050	139 264 515	Sleep duration	0.005	0.001	0.954	0.040
11	55 082 693	58 455 737	Ease of getting up	0.028	0.001	0.532	0.440
11	55 082 693	58 455 737	Morningness	0.021	0.000	0.738	0.241
12	46 024 786	47 713 678	Ease of getting up	0.001	0.005	0.814	0.180
12	46 024 786	47 713 678	Morningness	0.001	0.000	0.824	0.174
12	101 862 690	102 963 962	OSA	0.001	0.209	0.629	0.155
12	101 862 690	102 964 280	Snoring	0.000	0.372	0.527	0.092
17	43 056 905	45 874 355	Ease of getting up	0.000	0.005	0.942	0.053
17	43 056 905	45 874 355	Napping	0.000	0.000	0.984	0.016

Genomic coordinates are based on the human genome build GRCh37/hg19. PPA stands for Posterior Probability of Association. Models 1–4 represent the probability of the variant being causal for PD only (Model 1), the sleep trait only (Model 2), shared (Model 3), or distinct (Model 4).

Figure 2.

Shared genomic segments and genes between PD and sleep-related traits. Phenogram illustrating the chromosomal locations of genomic segments showing evidence of shared causal variants between PD and individual sleep-related traits identified via GWAS-pairwise (PPA > 0.5 for Model 3; see Methods). Genes located within each shared segment are annotated, color-coded per sleep-related trait.

Multiple genes located on chromosome 17 (WNT3, CRHR1, KANSL1, and MAPT) were among the most significantly associated with PD, ease of getting up, and napping. Shared loci between PD and ease of getting up were also observed on chromosomes 5, 11, and 12, whereas loci shared between PD and napping were restricted to chromosome 17. All loci associated with both PD and sleep duration were located on chromosome five. Notably, chromosomes 5, 12, and 17 contained the top genes associated with PD (Table 2).

Table 2.

Biological relevance of the protein-coding genes associated with PD and at least one sleep-related trait

Shared gene	Locus	PD p-value	Sleep-related trait p-value	Biological relevance
*WNT3	17q21.31	1.18E-06	Getting-up: 2.20E-07; napping: 8.90E-12	Encodes WNT3 protein, a key actor in Wnt signaling and development
*CRHR1	17q21.31	3.75E-07	Getting-up: 1.00E-08; napping: 1.15E-12	G-protein-coupled receptor for CRH; plays a key role in HPA stress responses
*KANSL1	17q21.31	3.22E-15	Getting-up: 9.89E-09; napping: 4.90E-12	Part of NSL complex; chromatin regulation and gene expression
*MAPT	17q21.31	1.10E-04	Getting-up: 4.78E-09; napping: 7.40E-13	Tau; stabilizes microtubular structures, neuronal polarity, synaptic maintenance
*NSF	17q21.31	5.00E-10	Getting-up: 1.08E-07; napping: 4.98E-11	ATPase for membrane fusion; synaptic vesicle release and recycling
*PLEKHM1	17q21.31	1.39E-14	Getting-up: 4.55E-08; napping: 5.44E-12	Adaptor for autophagosome-lysosome fusion; Rab7-dependent
*SCAF11	12q12	5.00E-10	Getting-up: 5.54E-07; morningness: 5.00E-10	Pre-mRNA splicing; spliceosome assembly
*C5orf30	5q21.1	5.55E-17	Sleep-duration: 1.45E-06	Negatively regulate macrophage resolution of inflammation; MACIR
*ARHGAP27	17q21.31	1.49E-06	Getting-up: 1.37E-07; napping: 5.92E-12	Rho GTPase-activating; involved in clathrin-mediated endocytosis
*PAM	5q21.1	4.48E-07	Sleep-duration: 2.12E-12	Peptidylglycine α-amidating; neuropeptide maturation
*ARID2	12q12	3.30E-04	Getting-up: 2.12E-08; morningness: 4.81E-10	Chromatin remodeler; part of PBAF complex
*GIN1	5q21.1	4.31E-05	Sleep-duration: 2.09E-09	Gypsy retrotransposon integrase 1; possibly involved in metabolic regulation
SYCP3	12q23.2	3.05E-04	OSA: 5.50E-04	Synaptonemal complex component; required for homologous pairing in meiosis
DRAM1	12q23.2	5.97E-06	OSA: 1.34E-06; snoring: 5.59E-08	Lysosomal membrane protein; autophagy, stress responses
ZFP91	11q12.1	1.58E-05	Getting-up: 1.30E-04; morningness: 2.59E-09	E3 ligase; non-canonical NF-κB signaling, cell proliferation
CNTF	11q12.1	1.39E-05	Morningness: 1.13E-04	Neurotrophic cytokine; neuronal survival, differentiation, repair
LPXN	11q12.1	1.39E-05	Morningness: 3.87E-07	Leupaxin; focal adhesion, cell migration, integrin signaling
*PPIP5K2	5q21.1	4.69E-05	Sleep-duration: 1.77E-06	Kinase; phosphoregulates inositol pyrophosphate signals
GLYAT	11q12.1	1.27E-04	Morningness: 1.12E-04	Detoxification enzyme; converts xenobiotics to acylglycines
GNPTAB	12q23.2	1.21E-06	OSA: 2.30E-04	Golgi enzyme; phosphotransferase adding mannose-6-P signals for delivery to lysosome
C11orf31	11q12.1	1.31E-04	Getting-up: 3.62E-07	Selenoprotein H; redox regulation, antioxidant, neuroprotection
*SPPL2C	17q21.31	1.37E-05	Getting-up: 2.05E-08; napping: 1.86E-12	Signal Peptide Peptidase-Like 2C; intramembrane protease
TMEM161B	5q14.3	8.07E-08	Getting-up: 1.94E-09; sleep-duration: 1.17E-04	Transmembrane 161B; involved in neuronal and cardiac development
CHPT1	12q23.2	1.87E-05	OSA: 3.16E-04	Choline phosphotransferase; phosphatidylcholine biosynthesis, membrane integrity
OR5B2	11q12.1	6.77E-07	Morningness: 3.15E-09	Olfactory receptor; G-protein-coupled receptor for odorants

For each gene, the chromosomal locus, the gene-based p-value for PD, the associated sleep-related traits, their corresponding p-values, and the number of contributing single-nucleotide polymorphisms are listed. Genes marked with an asterisk (^*) are top-ranked for PD based on gene-based test significance. Biological relevance is summarized based on published functional annotations.

We next performed gene-set enrichment analysis using MAGMA for PD and the self-reported sleep-related traits that showed the greatest number of shared genomic segments with PD: ease of getting up, napping, and sleep duration. For PD, significant enrichment was observed for pathways involved in sphingolipid metabolism, microglial proliferation, miRNA regulation, and the PI3K-AKT signaling pathway (Supplementary Table S6). For ease of getting up, enriched gene sets included those related to transcriptional regulation, circadian rhythms, synaptic signaling, and neurotransmission (Supplementary Table S7). In the case of napping, we identified enriched pathways involved in dopaminergic neurotransmission, sleep–wake cycle regulation, neuronal lipid metabolism, intracellular signaling cascades, and inflammatory and immune-related processes (Supplementary Table S8). For sleep duration, enriched gene sets were related to neuronal function, synaptic plasticity, neurotransmission, and signal processing (Supplementary Table S9).

The PPI network generated using STRING for genes mapped to the 17q21.31 locus indicated modules enriched for traits related to brain structure, mood and affective symptoms, and sleep-related traits (Supplementary Table S10). Genes such as MAPT, CRHR1, and KANSL1 formed high-confidence interactions within the network (25 nodes, 20 edges; expected = 1; PPI enrichment p < 1.0e−16), suggesting greater connectivity than expected by chance (Figure 3). Using the MCL approach, we identified four distinct clusters. The largest cluster included genes from the 17q21.31 region, ARHGAP27, CRHR1, MAPT, KANSL1, PLEKHM1, SPPL2C, and WNT3, while the remaining clusters grouped genes, suggesting a shared functional role or a similar pattern of connections in the protein–protein interaction map.

Figure 3.

Protein–protein interaction network of genes located in shared genomic segments. The network was generated using STRING and includes all significant protein-coding genes mapped to the shared loci identified via GWAS-PW. Functional modules were identified using the MCL (inflation parameter = 3). The resulting clusters correspond to distinct chromosomal loci, as indicated by the color-coded figure key. Edges represent protein–protein associations. Line colors indicate the primary evidence type supporting the interaction: black edges represent co-expression, and green edges indicate associations supported by text mining. Only moderate and high-confidence interactions (score ≥ 0.4) are shown. Full details on the evidence sources (including experimental and database scores) for each interaction are provided in Supplementary Table S11.

Discussion

We investigated the shared genetic architecture between PD and self-reported sleep-related traits by integrating GWAS summary statistics, conducting genetic correlation analyses, and mapping common causal variants to protein-coding genes. Although we did not detect statistically significant genome-wide genetic correlations between PD and most sleep-related traits after correcting for multiple comparisons, the GWAS-PW approach identified six genomic segments with evidence of shared causal variants. These segments were located on chromosomes 5, 11, 12, and, most prominently, chromosome 17, where the 17q21.31 locus contained the strongest associations.

In the present study, we did not observe statistically significant genome-wide genetic correlations between PD and the sleep traits after multiple testing correction, yet GWAS-PW identified several specific genomic segments that harbor shared causal variants. These results are not contradictory. Specifically, LDSC estimates the genome-wide alignment of effect sizes across genome-wide independent single-nucleotide polymorphisms. Therefore, if shared biology is restricted to a small subset of loci, or when shared loci include effects in opposite directions across traits, the resulting genome-wide correlation can be close to zero [21]. In contrast, GWAS-PW focuses on a level of LD-defined genomic segments and is designed to detect local pleiotropy even if genome-wide covariance is minimal. Thus, our findings suggest that the genetic overlap between PD and specific sleep traits could be concentrated in specific genomic segments rather than at the whole-genome level, with loci such as 17q21.31 prominently contributing to shared biological pathways.

Notably, the strength of evidence varied across these regions (Table 1). The 17q21.31 locus exhibited the highest posterior probability for a shared causal variant (PPA > 0.94), indicating high confidence in a shared genetic etiology. In contrast, shared loci on chromosomes 5 and 11 showed more moderate evidence (PPA ~0.50–0.53), suggesting that for these regions, the presence of distinct causal variants (Model 4) remains a relevant competing hypothesis.

Although some variability was observed in the strength of evidence across genomic segments prioritized by GWAS-PW, the majority showed strong support for a shared causal model. Specifically, most loci exhibited posterior probabilities for model 3 (shared association) > 0.73 (Table 1). The strongest evidence was observed at the 17q21.31 locus, which showed consistently high posterior probabilities for a shared causal variant across multiple sleep traits (PPA_Model3 > 0.94), indicating high confidence in shared genetic etiology between PD and sleep-related phenotypes. In contrast, genomic segments on chromosomes 5 and 11 displayed a more modest, though still supportive, evidence for shared association (PPA_Model3 ≈ 0.53–0.74). Therefore, our results suggest that there is shared genetic liability between PD and sleep-related traits, with a small number of loci, most notably 17q21.31, making most of the contributions to the observed overlap.

Our results suggest that the relationship between PD and sleep disturbances may be mediated through specific biological axes. First, genes involved in synaptic structure, neurotransmission, and neuronal survival, such as MAPT, NSF, ARHGAP27, and CNTF, play roles in axonal transport, vesicle trafficking, and synaptic regulation [31–35]. Second, genes related to circadian and neuroendocrine regulation, including CRHR1, ARHGAP27, WNT3, and OR5B2 are implicated in sleep timing, hypothalamic–pituitary–adrenal (HPA) axis activation, and circadian gene expression [36–40]. Third, autophagy- and lysosomal-related genes, such as PLEKHM1, DRAM1, and GNPTAB, are key to neuronal homeostasis and protein clearance, while CHPT1 supports the membrane integrity required for these processes [41–46]. Fourth, we identified a cluster of genes involved in metabolic regulation and redox homeostasis, specifically GLYAT, C11orf31, and PPIP5K2, which point towards detoxification and oxidative stress responses as shared mechanisms [47–52]. Fifth, transcriptional and post-transcriptional regulators, including ARID2, SCAF11, and KANSL1, implicate chromatin remodeling and splicing in the maintenance of dopaminergic and sleep-regulatory circuits [53–55]. Notably, ARHGAP27 and CRHR1 are involved in both neurotransmission and circadian regulation, suggesting that convergence across these biological domains may play an essential role in linking PD to sleep-related phenotypes [37].

The 17q21.31 locus has previously been associated with PD risk [56, 57]. In our study, we identified multiple genes within this region, including CRHR1, MAPT, KANSL1, NSF, WNT3, PLEKHM1, and ARHGAP27, that were significantly associated with both PD and sleep-related traits, such as ease of getting up and napping. Additionally, their high degree of connectivity in the PPI network suggests coordinated regulation and shared participation in functional pathways relevant to both sleep and neurodegeneration.

MAPT and KANSL1, located within the 17q21.31 inversion polymorphism, have been co-implicated in PD risk via sub-haplotypes that modulate gene expression [57]. Genes prioritized in our study, including MAPT, KANSL1, NSF, and ARHGAP27, are central to pathways supporting synaptic integrity. Specifically, MAPT encodes the tau protein, crucial for maintaining axonal structure and transport [58], while NSF is essential for vesicle fusion and neurotransmitter release [31], and ARHGAP27, a Rho GTPase-activating protein, regulates cytoskeletal dynamics and neuronal signaling [33]. By sustaining synaptic stability, MAPT, KANSL1, NSF, and ARHGAP27 may also contribute to the neural circuits underpinning circadian and sleep–wake processes, providing a plausible pathway through which genetic variation in these loci could contribute to the modulation of sleep-related traits [59].

CRHR1 encodes the corticotropin-releasing hormone receptor 1, a central regulator of the HPA axis and circadian rhythms [60, 61]. Its expression has been linked to sleep timing, duration, mood regulation, and stress reactivity [62]. Chronic HPA axis hyperactivity has been associated with both PD progression and sleep disturbances, positioning CRHR1 as a potential mechanistic link between these phenotypes, likely through glucocorticoid-mediated modulation of dopaminergic and circadian circuits [63, 64]. Similarly, WNT3, a component of the Wnt signaling pathway, is involved in neurodevelopment and synaptic plasticity, which may also be relevant to sleep–brain interactions through the regulation of adult neurogenesis and synaptic remodeling processes implicated in PD and sleep regulation [65, 66].

PLEKHM1, implicated in lysosomal trafficking and autophagy, also emerged as a candidate gene of interest. Autophagy dysfunction is a recognized hallmark of PD and has been linked to disrupted sleep regulation via circadian rhythm dysregulation [67–70]. Impairment in lysosomal clearance may lead to alpha-synuclein accumulation and neuroinflammation, key mechanisms in PD pathogenesis that are also modulated by sleep [71]. Some of the enriched pathways in our study, including PI3K-AKT signaling, sphingolipid metabolism, and circadian regulation, have also been shown to involve astrocytic activity [72–74].

Beyond the prominent associations in the 17q21.31 locus, our analysis identified shared genomic segments on chromosomes 5, 11 and 12 that provide insights for the genomic architecture of the previously defined biological axes implicated in the comorbidity of PD and sleep disturbances. At the 5q21.1 locus, the co-localization of PAM with the inflammatory regulator C5orf30 and the kinase PPIP5K2 highlights a convergence of neuropeptide maturation and neuro-immune pathways, whereas the 5q14.3 locus implicates the signaling gene TMEM161B [75–77]. Similarly, the 12q12 locus groups regulators of gene expression, including ARID2 and SCAF11, while the 12q23.2 locus is particularly notable for harboring genes involved in membrane dynamics and autophagy, such as DRAM1, GNPTAB, and CHPT1, suggesting a potential genomic regulatory block essential for lysosomal maintenance. Finally, the 11q12.1 region integrates the neurotrophic factor CNTF with genes involved in cellular stress responses, including ZFP91 and LPXN, providing a distinct neuroprotective target beyond the classic tau-related pathways [78, 79].

Gene-set enrichment analyses revealed convergence on pathways involved in synaptic and neurotransmission functions across PD, ease of getting up, and sleep duration, highlighting genes such as NSF, MAPT, and ARHGAP27. In addition, gene sets associated with dopaminergic and adrenergic signaling were significantly enriched in comparisons involving sleep duration. These findings suggest that shared genetic mechanisms may involve both neurotransmission and circadian processes.

We also identified enrichment of a gene set responsible for calcium-calcineurin (PPP3CA)-dependent signaling in both PD and sleep duration. This signaling axis, which is associated with sleep homeostasis and neurodevelopmental disorders involving seizures [80, 81], is active in neurons where it regulates synaptic vesicle recycling, dendritic plasticity, and aspects of circadian/sleep homeostasis [81]. Thus, the shared calcium-calcineurin-enriched gene set between PD and sleep duration may point towards a convergent mechanism involving proteostasis and vesicular trafficking dysfunction.

Several gene sets were shared across ease of getting up, napping, and sleep duration. For example, pathways related to sleep regulation were enriched for both ease of getting up and napping, consistent with gene-level findings implicating CRHR1 in circadian and neuroendocrine regulation [82]. The overlap of pathways involved in circadian regulation, synaptic function, and HPA axis activity suggests a biological basis for the relationship between sleep traits and vulnerability to neurodegeneration. Mechanistically, circadian disruption may impair dopaminergic neuron resilience, while alterations to synaptic plasticity and stress hormone dysregulation could exacerbate PD-related neurodegeneration [83–85]. These results align with the hypothesis of a bidirectional association between PD and sleep disturbances, in which shared genetic factors contribute to both neurodegenerative risk and susceptibility to sleep dysregulation [86].

There are several limitations to our study. First, the GWAS summary statistics were restricted to individuals of European ancestry, which limits the generalizability of our findings to other populations. Related to this, the present study relies on self-reported sleep phenotypes, which are justified in genomic research to maximize sample size and reduce the burden of in-depth interviews on participants [87]. Although subject to subjective bias, these traits currently represent the largest and most statistically powered GWAS datasets publicly available for sleep behaviors (n > 300 000), enabling reliable estimation of genetic correlations and identification of shared loci. Objective actigraphy-derived sleep measures, while valuable, are available only in the ∼90 000 participants of the UK Biobank subsample, and genome-wide analyses of these measures have smaller effective sample sizes and limited public availability of summary statistics. In addition, there are no GWAS available for other PD objective assessments, such as the Non-Motor Symptoms Questionnaire, which capture disturbances more precisely within the context of PD. As a result, incorporating actigraphy-based traits or PD objective assessments was not feasible in this study. Future work using large-scale actigraphy GWAS will be essential to evaluate whether the shared genetic signals identified here extend to objectively measured sleep architecture.

Second, while the 17q21.31 locus presents the strongest associations in the GWAS-PW analysis, interpreting individual causality at the individual gene level is constrained by its genetic architecture. The 17q21.31 locus is characterized by a large (~900 kb) inversion polymorphism and extensive linkage disequilibrium. Consequently, the association signal spans multiple co-inherited genes, making it challenging to statistically identify the specific driver. Therefore, the discussed genes by their functional relevance in this locus, such as MAPT, CRHR1, and KANSL1, should be considered as a candidate haplotype block rather than as confirmed individual causal genes [88].

Third, while gene-set enrichment analysis provides biological context, its interpretability is constrained by current limitations in gene-set annotation, and some pathways may be incompletely represented [89]. We also note that potential sample overlap between the PD and sleep GWAS datasets is unlikely to bias our results. Although the UK Biobank contributed to both the PD GWAS meta-analysis and to the sleep-related traits GWAS, this contribution was restricted to proxy cases in the PD GWAS meta-analysis (ie. participants who do not have the disease themselves but reported a first-degree relative with PD). Moreover, our genetic correlation analyses were conducted using LD Score Regression, whose intercept term accounts for sample overlap and prevents inflation of genetic covariance estimates [21]. GWAS-PW is also relatively robust to shared samples, as it evaluates the pattern of association across LD-defined genomic segments rather than relying on genome-wide covariance structures [22]. Together, these considerations suggest that the minimal sample overlap is unlikely to result in spurious associations. Lastly, we acknowledge the absence of certain PD-relevant sleep phenotypes such as RBD and RLS. Although GWAS studies for these traits have been published, the data may not be fully publicly available or the sample size may be too small to conduct post-GWAS analyses [90, 91]. As larger full-summary GWAS for these phenotypes become available, future work will be able to evaluate whether they exhibit stronger or additional shared genetic loci with PD beyond the traits examined here.

In summary, we provide evidence for shared genetic loci between PD and multiple self-reported sleep-related traits, particularly within the 17q21.31 region. This locus contains several genes with roles in neurodegeneration, circadian regulation, and stress response. Our findings support the hypothesis that sleep disturbances in PD may reflect not only disease progression but also shared etiological pathways. Converging results from gene-set analyses suggest that neurotransmission, circadian regulation, and neuroinflammatory processes are potential mechanisms underlying this overlap. These findings underscore the relevance of sleep-related endophenotypes in the genetic architecture of PD, suggesting new directions for early detection and mechanistic studies.

Author contributions

Aura Aguilar-Roldán (Formal analysis [lead], Writing—original draft [lead]), Natalia S. Ogonowski (Conceptualization [equal], Formal analysis [equal], Methodology [equal]), Miguel E. Rentería (Conceptualization [equal], Methodology [equal], Resources [lead], Supervision [equal], Writing—review & editing [equal]), and Luis M. Garcia Marin (Conceptualization [equal], Methodology [lead], Supervision [lead], Writing—original draft [equal], Writing—review & editing [equal])

Funding

This research was supported by the computational resources and infrastructure provided by the QIMR Berghofer. We also acknowledge the funding sources that supported the generation of the summary statistics used in this study, including the International Parkinson’s Disease Genomics Consortium (IPDGC), 23andMe and the UK Biobank.

Disclosure statement

Financial disclosure: None declared.

Non-financial disclosure: None declared.

Data availability

The full GWAS summary statistics for PD and sleep-related traits are available upon request through a Data Transfer Agreement and application procedure, including the 23andMe cohort (https://research.23andme.com/dataset-access/).

Code availability

No custom code was implemented in this study. All genetic analyses were performed using publicly available software, which is cited and described in the main text.

Ethics declaration

This study was conducted under the oversight of QIMR Berghofer’s Human Research Ethics Committee. All participants provided informed consent.

UK Biobank: The UK Biobank study was approved by the National Health Service National Research Ethics Service (ref. 11/NW/0382), and all participants gave their written consent to take part in the UK Biobank study. Further details about ethics oversight in UK Biobank can be found at https://www.ukbiobank.ac.uk/ethics/.

23andMe Inc: Participants provided consent and took part in the research online under a protocol approved by the independent AAHRPP-accredited Institutional Review Board, Ethical & Independent Review Services (E&I Review).